JP2014103486A - Image processing apparatus, image processing method, program and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obscure an interpolation pixel, by relaxing the discontinuity of pixel value of the interpolation pixel and a peripheral pixel.SOLUTION: A periodicity determination unit 101 determines whether or not there is the periodicity in variation of the pixel value in a predetermined area including an interpolation pixel. A first pixel value generation unit 106 generates a candidate pixel value of an interpolation pixel by a first interpolation method, and a second pixel value generation unit 107 generates a candidate pixel value of an interpolation pixel by a second interpolation method. When there is the periodicity, a control unit 103 selects the first pixel value generation unit 106, otherwise selects the second pixel value generation unit 107. An interpolation pixel value insertion unit 104 inserts a candidate pixel value of an interpolation pixel at the position of an interpolation pixel, and a local pixel value update unit 105 updates the pixel value of a local area, including an interpolation pixel and neighboring pixels, by using a predetermined correction amount.

Description

  The present invention relates to an image processing apparatus, an image processing method, a program, and a recording medium that perform interpolation processing on an image.

  In a reading unit such as a scanner that reads an image using optical means, the image signal may be lost or read due to a defect in the sensor that reads the image signal or due to the presence of a shielding object in the optical path such as dirt on the contact glass. The image signal may show an incorrect value. In addition, in a contact image sensor or the like, since a reading unit is configured by combining a plurality of short sensor chips, there is a case where a gap occurs between the sensor chips and an image signal cannot be acquired. An image read under such conditions deteriorates in image quality due to missing or incorrect values of some pixels.

  In order to improve this problem, there is an interpolation method that estimates and replaces pixel values of missing pixels or pixels having incorrect values using surrounding pixels. For example, in Patent Document 1, in order to accurately estimate the pixel value of an interpolation pixel, if the interpolation pixel position is within a halftone dot region, a pattern similar to the pattern including the interpolation pixel is searched from the image using pattern matching. Then, the pixel value of the pixel corresponding to the interpolation pixel in the most similar pattern is determined as the pixel value of the interpolation pixel.

  The conventional interpolation method has a problem in that the discontinuity of the pixel value between the replaced pixel and the surrounding pixels becomes conspicuous if the estimation accuracy of the pixel value is not sufficient. In particular, in a contact image sensor or the like, the gap width between sensor chips may not be an integral multiple of the pixel width due to assembly errors. Even for a digital image read using such a sensor, pixel values obtained from each sensor chip are generally arranged assuming that the gap width between sensor chips is an integral multiple of the pixel width. There are many cases. In such a case, even in the same digital image, distortion occurs between an image area (hereinafter referred to as a column) read by one sensor chip and an image area read by another sensor chip. Become.

  As an example of this image distortion, there is a case of reading a document on which parallel lines extending in the sub-scanning direction are drawn. In this case, if there is a gap between the sensor chips between the two straight lines, the distance between the two straight lines will be different from the distance between the other straight lines. In addition, when a document with a single straight line drawn obliquely is read by the above-mentioned sensor, the straight line seems to have been translated horizontally, vertically, and horizontally for each column of the read digital image due to sensor chip assembly errors. Will be configured.

  Since the above example relates to straight lines and parallel lines, sensor chip assembly errors rarely have a serious effect on image quality. This is because the number of gaps between sensor chips in a single sensor is not as large as the left, so that distortion occurring on a straight line is difficult to find. Further, regarding the problem that the interval between the parallel lines is different, the interval between the parallel lines is originally sufficiently wider than the assembly error of the sensor chip, so that it is difficult to make a perceivable difference.

  However, a halftone image is an example of a case where image quality is seriously affected by a problem similar to the above example. In the halftone image, the halftone dots are arranged in a straight line having a certain angle (screen angle) from the horizontal with a predetermined period (number of screen lines). The number of screen lines is generally 40 to 200 lines per inch. Especially in high-line images, the intervals between halftone dots are very close to about 0.1 mm. It may have a non-negligible effect on image quality. Furthermore, in the case where the gap between the sensor chips covers halftone dots, the shape of the halftone dots is distorted due to an assembly error, and it is difficult to estimate the pixel values of pixels (interpolated pixels) that are missing due to the gap. This is because the pixel value of the interpolation pixel (interpolation pixel value) is estimated using the left column as a reference and the result is different between the case where the interpolation pixel value is estimated using the right column as a reference. This is because the correct value does not exist in the pixel value filling the pixel. Even if measures are taken such that the interpolated pixel values obtained from the left and right columns are taken, discontinuity of pixel values across missing pixels due to image distortion may be perceived.

The present invention has been made in view of the above problems,
An object of the present invention is to reduce the discontinuity of pixel values between an interpolated pixel and a peripheral pixel by changing the pixel values of pixels existing around the interpolated pixel based on a predetermined standard, thereby making the interpolated pixel stand out. An object is to provide a lost image processing apparatus, an image processing method, a program, and a recording medium.

  According to the present invention, in an image processing apparatus that generates a pixel value of an interpolation pixel in a predetermined image by a predetermined interpolation method, whether or not the image region including the interpolation pixel is a periodic region with periodic pixel value fluctuations. Determining means for determining whether or not, first generation means for generating candidate pixel values of the interpolation pixels of the image using the first interpolation method, and candidate for interpolation pixels of the image using the second interpolation method When it is determined that the second generation unit that generates a pixel value and the determination unit are periodic regions, the candidate pixel value generated by the first generation unit is selected, and when the pixel value is determined not to be a periodic region, Control means for selecting a candidate pixel value generated by the second generation means; insertion means for inserting the selected candidate pixel value into the interpolation pixel; and the candidate pixel value of the interpolation pixel and the vicinity of the interpolation pixel. The pixel values of existing neighboring pixels are The most important further comprising a updating means for updating using the correction amount.

  According to the present invention, by changing the pixel values of the pixels existing around the interpolation pixel based on a predetermined criterion, the discontinuity of the pixel values of the interpolation pixel and the peripheral pixels is alleviated, and the interpolation pixel is made conspicuous Can be eliminated.

1 shows a configuration of an image processing apparatus according to an embodiment of the present invention. The processing flowchart of Example 1, 2 of this invention is shown. A detailed process flowchart of step 202, steps 302 to 304, and step 205 of the first embodiment is shown. An example of a signal observation position and a reproduced signal is shown. An example for explaining the update of the pixel value is shown. 10 is a detailed process flowchart of Step 205 according to the second embodiment.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present invention, during interpolation processing for interpolating information on missing pixels (interpolated pixels) in an image, pixel values of missing pixels in an input image are estimated and interpolated, and pixel values of missing pixels and missing pixels are interpolated. By changing the pixel values of the neighboring pixels based on a predetermined criterion, the discontinuity between the interpolated pixels and the neighboring pixels is relaxed, and a higher quality interpolation result is obtained.

  In the first embodiment, candidate pixel values of an interpolation pixel are generated using two pixel value generation units, the pixel value is inserted into the interpolation pixel, and the pixel value of the interpolation pixel and the pixel existing around the interpolation pixel It is an Example which changes a value based on a predetermined reference | standard.

  FIG. 1 shows a configuration of an image processing apparatus according to a first embodiment of the present invention. In FIG. 1, reference numeral 101 denotes a periodicity determination unit that determines whether or not a change in pixel value has periodicity in an image region having a predetermined size including an interpolation pixel, and reference numeral 102 denotes a periodic region and an aperiodic region. , And a boundary determination unit 103 for determining which region belongs to the boundary when it exists near the boundary. 103 is based on the determination results of the periodicity determination unit 101 and the boundary determination unit 102. A control unit that determines which one of the first pixel value generation unit 106 and the second pixel value generation unit 107 is used to generate a candidate pixel value, and 104 indicates a first pixel value generation unit 106 and a second pixel value generation unit An interpolation pixel value insertion unit that inserts the candidate pixel value of the interpolation pixel generated in any one of 107 at the position of the interpolation pixel, and 105 is based on a predetermined criterion for the pixel value of the local region including the interpolation pixel and the neighboring pixel Update local pixel value to update , 106 is a first pixel value generation unit that generates a candidate pixel value of an interpolation pixel using the first interpolation method, and 107 is a second pixel that generates a candidate pixel value of the interpolation pixel using the second interpolation method. It is a value generator.

  FIG. 2 shows a process flowchart of the first embodiment of the present invention.

(1) An interpolation pixel is set (S201).
The control unit 103 sets an interpolation pixel that is a target for inserting a pixel value from the processing target image. Interpolated pixels are given as isolated points or as a continuous line of points. As an example of a method for determining the position of the interpolation pixel, there are a method in which the user designates in advance, a detection in advance by the apparatus, and a sequential detection when setting the interpolation pixel. Examples of a method for detecting an interpolation pixel include a method of finding a pixel of brightness and color designated in advance, a method of evaluating the amount of deviation from a correct value when a known image is read, horizontal or There is a method of detecting a position where the periodicity in the vertical direction becomes discontinuous.

(2) The periodicity of the area including the interpolation pixel is determined (S202).
The periodicity determination unit 101 determines whether or not the variation of the pixel value has periodicity within a predetermined size area (determination area) including the interpolation pixel of the processing target image. FIG. 3A shows a detailed flowchart of step 202. First, the representative period is estimated (S301), and then the periodicity of the entire region is determined (S302). Thereafter, the region is divided into the left and right sides of the interpolation pixel, and after determining the periodicity in the left partial region (S303), the periodicity is determined in the right partial region (S304). Note that the left and right partial areas do not necessarily have to be divided areas.

1) Determination method of area size The determination area may be an area with a height of 1 pixel (one line of an image) including an interpolation pixel, or a rectangular area with a certain height including an interpolation pixel. Also good. The size of the area may be given in advance by the user or may be determined dynamically by the apparatus. It should be noted that the determination area is preferably given in the shape of the upper and lower sides and the left and right sides with the interpolation pixel as the center.

2) Method for determining periodicity FIG. 3B shows a detailed flowchart of steps 302 to 304. As an example of a method for determining whether or not the variation of the pixel value has periodicity, the variation period of the pixel value in the determination region is obtained (S401), and the reliability of the cycle is calculated (S402). There is a method of judging the reliability by threshold processing (S403).

  It should be noted that the step of estimating the representative period (S301) and the step of determining the entire periodicity (S302) may be shared with the step of estimating the period of the pixel value (S401). Further, the order of the steps for determining the periodicity (S302 to S304) is not limited, and any steps that can be shared may be combined as appropriate. For example, a flow in which the left and right periodicities are first determined and the overall periodicity is determined using the result is also conceivable.

a) As an example of a method for obtaining the representative period, a position (peak position) where the pixel value is maximum in the determination area is recorded, and a histogram of the distance from one peak position to the next peak position is recorded over the entire determination area. There is a method in which the distance that is the mode value is adopted as the representative period after the creation. In addition, you may use the position which becomes the minimum instead of the position which becomes maximum. However, since these methods using pixel values directly have low noise resistance, a method using autocorrelation is recommended.

  The autocorrelation is a correlation between a certain signal and a signal obtained by giving a predetermined phase shift to the signal. The autocorrelation is obtained within the determination region, the above-described histogram is created using the autocorrelation value instead of the pixel value, and the mode value is adopted as the representative period. As the autocorrelation value, a correlation coefficient may be used, or a covariance value may be substituted for easier calculation.

b) As an example of a method for obtaining the reliability, there is a method of multiplying the representative period by the frequency of the histogram of the distance and dividing by the number of pixels in the determination region. In other words, how much the distance between the peak positions that is the same as the representative period occupies the entire determination region is defined as the reliability. In addition, from the viewpoint of allowing a certain amount of error in the representative cycle, the frequency of the distance of the representative cycle ± 1 may be used in total.

c) Threshold processing of reliability If the reliability is higher than the threshold, it is determined that there is periodicity, otherwise it is determined that there is no periodicity. The reliability threshold value may be given in advance by the user, or may be dynamically determined by the present apparatus.

(3) Determining the boundary of the area including the interpolation pixel (S203)
The boundary determination unit 102 determines whether the interpolation pixel exists in the vicinity of the boundary between the periodic region and the non-periodic region, and determines whether the interpolation pixel belongs to the boundary pixel when it exists in the vicinity of the boundary.

1) Boundary determination method As an example of a method for determining whether or not a boundary exists, a reference area having a predetermined size is set at a position left and right by a predetermined distance from an interpolation pixel, and left and right reference is performed. There is a method in which the variance of pixel values is obtained individually in each region, and if only one of the left and right variances is greater than or equal to a threshold value, it is determined that the interpolated pixel exists in the vicinity of the boundary.

  In addition, as an example of a method for determining whether an interpolation pixel belongs to a periodic region or a non-periodic region, a reference region is set on the left and right of the interpolation pixel, and pixel value dispersion is separately obtained in the left and right reference regions. There is a method of determining that both the left and right sides exist on the region side with periodicity if the variance is equal to or greater than the threshold value, and that the other side is on the region side without periodicity.

2) Determination method of reference area size The reference area may be a 1-pixel-high area (one line of an image) including an interpolation pixel, or a rectangular area having a certain height including an interpolation pixel. May be. The size of the area may be given in advance by the user or may be determined dynamically by the apparatus. Further, the variance may be replaced by the difference between the maximum luminance value and the minimum luminance value of the pixels belonging to the reference area.

(4) An interpolation method used for pixel interpolation is determined (S204).
Based on the results of the periodicity determination unit 101 and the boundary determination unit 102, the control unit 103 generates a candidate pixel value of the interpolation pixel using either the first pixel value generation unit 106 or the second pixel value generation unit 107. Determine whether. As an example of an interpolation method determination method, the first pixel value generation unit 106 is selected when the periodicity determination unit 101 determines that there is periodicity, and the second pixel value generation unit 107 is selected otherwise. The method of doing is mentioned.

  Even if the periodicity determination unit 101 determines that the region including the interpolation pixel has periodicity, the second pixel is detected when the boundary determination unit 102 determines that the interpolation pixel exists in the non-periodic region. There is also a method for selecting the value generation unit 107. If the first pixel value generation unit 106 is selected, the process proceeds to S205. If the second pixel value generation unit 107 is selected, the process proceeds to S206.

(5) Generating candidate pixel values using the first pixel value generating unit 106 (S205)
The first pixel value generation unit 106 generates candidate pixel values for interpolation pixels. An example of the interpolation method used here is a pattern matching method. A specific example of the pattern matching method is template matching. As a scale for measuring the degree of coincidence between patterns used in template matching, a degree of difference such as SAD or SSD, or a degree of similarity such as a correlation coefficient or covariance can be considered. An example of a procedure for generating a pixel value using template matching is as follows. FIG. 3C shows a detailed processing flowchart of step 205.

1) Determine the size of the template (S501)
The width of the template is set based on the representative period. The width of the template may be matched with the representative period, or may be set slightly larger or smaller than the representative period. The height of the template may be matched with the representative period, or may be another value.

  When the interpolation pixel belongs to the halftone dot area, it is desirable to set it as a square area whose one side is approximately the representative period. Thereby, when searching for a similar pattern, only information for one halftone dot can be used, and a similar pattern optimum for a halftone dot including an interpolation pixel can be acquired. When the representative period is very small (such as 3 pixels), the template size may be slightly larger (such as 5 pixels). This is because if the representative period is very small, the number of pixels used for the evaluation of coincidence decreases, and the stability of similar pattern detection decreases.

2) Create a template (reference pattern) (S502)
A region having the size of the template determined in 1) including the interpolation pixel is cut out from the processing target image and set in the template. The template is preferably set in a shape that is symmetrical with respect to the left and right and up and down with the interpolation pixel as the center. This makes it possible to perform a search without depending on the directionality of the pattern including the interpolation pixel.

3) A search range is determined (S503).
In the processing target image, a search range that is a range for searching for a similar pattern is set. The width and height of the search range may be set based on the representative period estimated in S203, or may be another value. It is desirable that the search range has the same width and height. Thereby, it becomes possible to perform a search without depending on the directionality of the pattern in the search range. However, if the height is to be kept low due to restrictions on the memory for buffering images, it is desirable to increase the width accordingly.

  When the periodicity determination unit 101 determines that there is no periodicity on the left side, it is desirable to exclude the left side of the interpolation pixel from the search range. Similarly, when it is determined that there is no periodicity on the right side, it is desirable to exclude the right side of the interpolation pixel from the search range. By narrowing the search range in this way, erroneous detection from an area where a similar pattern should not be detected can be prevented.

4) Search for similar patterns (S504)
At each position within the search range determined in 3), the degree of coincidence with the template created in 2) is obtained, and the position with the highest degree of coincidence (low dissimilarity or high similarity) is selected as a similar pattern. . When evaluating the degree of coincidence, the interpolation pixel is excluded from the coincidence calculation so that the pixel value of the interpolation pixel does not affect the degree of coincidence, or a temporary value is used using the pixel value at the same position in the comparison target pattern. It is desirable to set a pixel value. A plurality of similar patterns may be selected in descending order of coincidence.

5) Obtaining candidate pixel values (S505)
In the similar pattern, a pixel value of a pixel (corresponding pixel) at a position corresponding to the position of the interpolation pixel in the template is acquired from the processing target image, and is set as a candidate pixel value of the interpolation pixel. When there are a plurality of similar patterns, a plurality of candidate pixel values are obtained according to the reference, and the pixel values are synthesized. As a synthesis method, averaging may be performed using uniform weights, or weighted averaging may be performed after controlling so that the higher the matching degree of similar patterns, the greater the weight.

(6) Generating candidate pixel values using the second pixel value generation unit (S206)
The second pixel value generation unit 107 generates candidate pixel values for interpolation pixels. An example of the interpolation method used here is an interpolation method. The interpolation method is a method of using peripheral pixel values when deriving a pixel value of a certain pixel. Examples of interpolation methods include nearest neighbor interpolation (zero order interpolation), linear interpolation (primary interpolation), parabolic interpolation (second order interpolation), cubic interpolation (third order interpolation), polynomial interpolation, spline interpolation, Lagrange interpolation, Etc. Further, bilinear interpolation or bicubic interpolation obtained by extending the above-described linear interpolation or cubic interpolation to two dimensions may be used.

(7) The candidate pixel value is inserted into the interpolation pixel (S207)
The interpolation pixel value insertion unit 104 inserts the candidate pixel value of the interpolation pixel generated in the previous step at the position of the interpolation pixel. Thereby, the interpolation process for one interpolation pixel is completed.

(8) Update the pixel values of the interpolated pixel and neighboring pixels (S208).
The local pixel value update unit 105 updates the pixel value of a local area including the interpolation pixel and a neighboring pixel located in the vicinity thereof based on a predetermined reference. Examples of the method for updating the pixel value include the following.

a) Update pixel values of neighboring pixels (interpolated pixel values remain unchanged)
This is a method of giving an offset to the pixel values of neighboring pixels so that the average pixel value of the local area approaches the average pixel value of an area of the same size (peripheral area) existing around the local area.

To select the surrounding area,
・ A method that adopts a position that is an integer multiple of the representative period vertically or horizontally from the local area ・ Set a local area as a template, and detect and adopt a pattern similar to the local area from a predetermined range by template matching The method etc. are mentioned. It should be noted that a plurality of peripheral regions may be selected, or may be selected one by one on the left and right or top and bottom when viewed from the local region.

Also, how to set the offset amount is
-A method of setting the same for all neighboring pixels-A method of setting so as to decrease as the distance from the interpolated pixel-A pixel adjacent to the interpolated pixel is relatively small, and then relatively large. The method of control etc. are mentioned. Note that the offset is set so that the average pixel value of the local region approaches the average pixel value of the peripheral region, and therefore, these do not necessarily match. In determining the offset amount, a weight amount given in advance may be used to determine how close the former is to the latter. Further, a range in which pixel values are updated may be a range given in advance.

b) Update the pixel values of the interpolated pixel and neighboring pixels (the interpolated pixel value also changes).
This is a method of giving an offset to the pixel values of both the interpolated pixel and the neighboring pixels so that the average pixel value of the local region approaches the average pixel value of the region of the same size (peripheral region) existing around the local region. . The method described in (a) above may be used as a peripheral region selection method and offset amount setting method.

  Regarding the method of setting the offset amount, examples of a method different from the above-described (a) include the following. First, the difference between the pixel value of the pixel (corresponding pixel) corresponding to the position of the interpolation pixel in the local region and the interpolation pixel value in the peripheral region is taken. This difference is given to the interpolated pixel value as an offset amount so as to match the pixel value of the corresponding pixel. Next, a value obtained by reversing the sign of the above-described difference is obtained, and this value is distributed to neighboring pixels to obtain the offset amount of neighboring pixels. The distribution method may be a method similar to the offset amount setting method described in (a) above, and the sum of the offset amounts of all neighboring pixels becomes equal to a value obtained by reversing the sign of the difference.

c) This is a method of smoothing fluctuations in pixel values by applying a moving average to pixel values of pixels existing in a local area to be updated by applying a moving average to pixel values of interpolated pixels and neighboring pixels. The average pixel value in the local area does not change, but the gradient becomes gentle.

d) Obtaining a weighted average of the pixel values of the local area and the peripheral area to be updated by combining the pixel values of the interpolated pixel and the neighboring pixels with the pixel values of the peripheral area, and updating the pixel values of the local area with this value It is a method to do. The method described in (a) above may be used as the peripheral region selection method. The weight used for the weighted average of the pixel values in the local area and the pixel values in the peripheral area may be changed for each neighboring pixel with reference to the offset amount setting method (a) described above.

  With the sensor chip assembly error, the width of the gap between the sensor chips increases or decreases. If the design value of the gap width is 1 pixel ± 0.5 pixel, the gap width is 0.5 pixels at the minimum and 1.5 pixels at the maximum. It is desirable to measure the gap width in advance, align the image areas read by the left and right sensor chips accordingly, and then create an image by arranging these image areas. However, a process for measuring the gap width is required, and not only processing for aligning the image area is required, but also the high-frequency component of the image may be dull depending on the alignment method. For these reasons, it is common to arrange the image regions without ignoring the above-mentioned alignment and ignoring the error. However, problems may arise with this response.

  The concept will be described with reference to FIG. In FIG. 4, a signal represented by a solid curve is sampled at a position represented by a straight line. Here, a straight line represented by a solid line is an ideal sampling position, and the gap width is 1 pixel if the center is a gap. Although a signal cannot be detected at the central sampling position corresponding to the gap, it is assumed here that the signal has been detected or appropriately interpolated in order to simplify the description. Similarly, in order to simplify the description, it is assumed that the assembly error is distributed symmetrically about the gap position.

  When the gap width is wider than one pixel, the sampling position is centered on the gap position rather than the ideal position, as shown by a one-dot chain line in FIG. When the signal observed in this way is reproduced assuming that the gap width is one pixel, the reproduced signal has a width as compared with the actual signal, as shown by a one-dot chain line in FIG. It narrows. Conversely, when the gap width is narrower than one pixel, the sampling position is closer to the center of the gap position than the ideal position as shown by the dotted line in FIG.

  When the signal observed in this way is reproduced assuming that the gap width is one pixel, the reproduced signal is wider than the actual signal as shown by the dotted line in FIG. 4B. . Assuming that the value of the signal at the gap position takes a true value, the local average value decreases as the signal width decreases, and conversely the local average value increases as the signal width increases. On the other hand, even when the value of the signal at the gap position deviates from the true value and the local average value coincides with the surrounding area, the fluctuation in the signal width is likely to be perceived as a local frequency fluctuation.

  From such a background, “a) Update the pixel value of the neighboring pixel” is performed by updating the pixel value of the neighboring pixel without changing it under the assumption that the pixel value of the interpolation pixel is a true value. It is intended to bring the reproduced signal closer to the actual signal. In addition, “b) Update the pixel values of the interpolation pixel and the neighboring pixel” is a pixel value from the pixel value of the interpolation pixel to the neighboring pixel under the assumption that the pixel value of the interpolation pixel is out of the true value. Is to disperse. “C) Update pixel values of interpolated pixels and neighboring pixels by moving average” is a method close to the latter, and is suitable for the case where the pixel value of a specific pixel is too large or too small. This error is suppressed by propagating to the surroundings. “D) Update the pixel value of the interpolated pixel and the neighboring pixel by combining with the pixel value of the peripheral region” is an intermediate method between the former and the latter.

  The case of “b) updating pixel values of interpolated pixels and neighboring pixels” will be further described with reference to FIG. FIG. 5A shows a case in which the average width is maintained while the signal width is narrowed, while the pixel value of the interpolation pixel is decreased, and an amount corresponding to the decrease is distributed to the neighboring pixels. By increasing the pixel value, the shape of the signal can be made closer to the actual one. FIG. 5B shows a case in which the width of the signal is widened, while the average value is kept small, and the pixel value of the interpolation pixel is increased, and the amount corresponding to the increased amount is set to the pixel value of the neighboring pixel. By distributing and reducing the pixel value, the shape of the signal can be made closer to the actual one. In “a) updating the pixel value of the neighboring pixel”, it can be considered that the similar processing is performed on the neighboring pixel without changing the pixel value of the interpolation pixel.

  (9) S202 to S208 are executed for all the interpolated pixels (S209). When there are a plurality of interpolation pixels, the processing from S202 to S208 is repeated by the number of interpolation pixels.

Example 2 is an example in which the offset amount (correction amount) of Example 1 is calculated based on a template and a similar pattern. The configuration and processing flowchart of the second embodiment are the same as those of the first embodiment. The difference from the first embodiment is the processing in step 205. FIG. 6 is a detailed processing flowchart of step 205 in the second embodiment.
(1) Set interpolation pixels (S201)
(2) The periodicity of the area including the interpolation pixel is determined (S202).
(3) Determining the boundary of the area including the interpolation pixel (S203)
(4) An interpolation method used for pixel interpolation is determined (S204).
(5) Generating candidate pixel values using the first pixel value generating unit (S205)
The first pixel value generation unit 106 generates candidate pixel values for interpolation pixels. An example of the interpolation method used here is a pattern matching method. A specific example of the pattern matching method is template matching. As a scale for measuring the degree of coincidence between patterns used in template matching, a degree of difference such as SAD or SSD, or a degree of similarity such as a correlation coefficient or covariance can be considered. An example of a procedure for generating a pixel value using template matching is as follows.

1) Determine the size of the template (S601)
The width of the template is set based on the representative period. The width of the template may be matched with the representative period, or may be set slightly larger or smaller than the representative period. The height of the template may be matched with the representative period, or may be another value.

  When the interpolation pixel belongs to the halftone dot area, it is desirable to set it as a square area whose one side is approximately the representative period. Thereby, when searching for a similar pattern, only information for one halftone dot can be used, and a similar pattern optimum for a halftone dot including an interpolation pixel can be acquired. When the representative period is very small (such as 3 pixels), the template size may be slightly larger (such as 5 pixels). This is because if the representative period is very small, the number of pixels used for the evaluation of coincidence decreases, and the stability of similar pattern detection decreases.

2) Create a template (reference pattern) (S602)
A region having the size of the template determined in 1) including the interpolation pixel is cut out from the processing target image and set in the template. The template is preferably set in a shape that is symmetrical with respect to the left and right and up and down with the interpolation pixel as the center. This makes it possible to perform a search without depending on the directionality of the pattern including the interpolation pixel.

3) A search range is determined (S603).
In the processing target image, a search range that is a range for searching for a similar pattern is set. The width and height of the search range may be set based on the representative period estimated in S203, or may be another value. It is desirable that the search range has the same width and height. Thereby, it becomes possible to perform a search without depending on the directionality of the pattern in the search range. However, if the height is to be kept low due to restrictions on the memory for buffering images, it is desirable to increase the width accordingly.

  When the periodicity determination unit 101 determines that there is no periodicity on the left side, it is desirable to exclude the left side of the interpolation pixel from the search range. Similarly, when it is determined that there is no periodicity on the right side, it is desirable to exclude the right side of the interpolation pixel from the search range. By narrowing the search range in this way, erroneous detection from an area where a similar pattern should not be detected can be prevented.

4) Search for similar patterns (S604)
At each position within the search range determined in 3), the degree of coincidence with the template created in 2) is obtained, and the position with the highest degree of coincidence (low dissimilarity or high similarity) is selected as a similar pattern. . When evaluating the degree of coincidence, the interpolation pixel is excluded from the coincidence calculation so that the pixel value of the interpolation pixel does not affect the degree of coincidence, or a temporary value is used using the pixel value at the same position in the comparison target pattern. It is desirable to set a pixel value. A plurality of similar patterns may be selected in descending order of coincidence.

5) Obtaining candidate pixel values (S605)
In the similar pattern, the pixel value of the pixel (corresponding pixel) at the position corresponding to the position of the interpolation pixel in the template is acquired from the processing target image and set as the candidate pixel value of the interpolation pixel. When there are a plurality of similar patterns, a plurality of candidate pixel values are obtained according to the reference, and the pixel values are synthesized. As a synthesis method, averaging may be performed using uniform weights, or weighted averaging may be performed after controlling so that the higher the matching degree of similar patterns, the greater the weight.

6) A correction amount of the candidate pixel value is calculated (S606).
Using the template and the similar pattern, a correction amount to be given to the candidate pixel value is calculated. Examples of the correction amount include the following.

a) Use the difference between the average pixel value of the template and the average pixel value of the similar pattern. b) The average value of the maximum pixel value and the minimum pixel value of the template, and the average value of the maximum pixel value and the minimum pixel value of the similar pattern. Use a difference c) When there are a plurality of similar patterns that use the difference between the average value of the pixel values of the left and right pixels of the interpolation pixel and the average value of the pixel values of the left and right pixels of the corresponding pixel, What is necessary is just to synthesize | combine the obtained correction amount. The combining method is the same as that used when combining a plurality of candidate pixel values.

(6) Generating candidate pixel values using the second pixel value generation unit (S206)
(7) The candidate pixel value is inserted into the interpolation pixel (S207)
(8) Update the pixel values of the interpolated pixel and neighboring pixels (S208).
The local pixel value updating unit 105 updates the pixel value of the local area including the interpolation pixel and the neighboring pixel located in the vicinity thereof based on a predetermined criterion using the correction amount. Examples of the method for updating the pixel value include the following.

a) This is a method of improving this by the correction amount under the assumption that the pixel value of the interpolation pixel that gives the correction amount to the pixel value of the interpolation pixel is out of the true value.

b) Giving correction values to the pixel values of the interpolated pixels and neighboring pixels In the method a), the pixel values of the interpolating pixels and the neighboring pixels may be discontinuous. And give it. As a method for setting the weight when the correction amount is distributed and given, the method described in the first embodiment may be used.

  The reason for giving the correction amount is that there is a case where a similar pattern that completely matches the template cannot be found. For example, it is assumed that the image has a locally dark area such as the shadow of a white panel on a white wall and the surrounding area is a relatively bright area, and the interpolation pixel exists in the shadow area. . If a similar pattern is found from a relatively bright area by pattern matching, if the candidate pixel value of the interpolation pixel is inserted into the interpolation pixel as it is, there is a risk of discontinuity with the pixel values of neighboring pixels. For this reason, it is necessary to ensure continuity with the pixel values of neighboring pixels by correcting the pixel values of the interpolated pixels after insertion. In addition, if only the pixel value of the interpolated pixel is corrected, there may be a case where sufficient continuity with the pixel value of the neighboring pixel cannot be ensured. Therefore, it is desirable to correct the pixel value of the neighboring pixel together.

  (9) S202 to S208 are executed for all the interpolated pixels (S209).

  As described above, according to the present invention, the pixel values of pixels (neighboring pixels) existing within a predetermined range from the interpolation pixel, and the pixel values from the further outer pixel (peripheral pixel) to the interpolation pixel are continuously obtained. To change to change (changes are easy to be perceived when fluctuation occurs at a short distance, but changes gradually become difficult to perceive when changing gradually over a long distance, so discontinuity is widened. Disperse), the discontinuity of the pixel values of the interpolated pixel and the surrounding pixels is alleviated, and the interpolated pixel becomes inconspicuous.

  According to the present invention, a storage medium storing software program codes for realizing the functions of the above-described embodiments is supplied to a system or apparatus, and the computer (CPU or MPU) of the system or apparatus is stored in the storage medium. It is also achieved by reading and executing the program code. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment. As a storage medium for supplying the program code, for example, a hard disk, an optical disk, a magneto-optical disk, a nonvolatile memory card, a ROM, or the like can be used. Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) operating on the computer based on an instruction of the program code. A case where part or all of the actual processing is performed and the functions of the above-described embodiments are realized by the processing is also included. Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. A case where the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing and the functions of the above-described embodiments are realized by the processing is included. Further, the program for realizing the functions and the like of the embodiments of the present invention may be provided from a server by communication via a network.

DESCRIPTION OF SYMBOLS 101 Periodicity determination part 102 Boundary determination part 103 Control part 104 Interpolation pixel value insertion part 105 Local pixel value update part 106 1st pixel value generation part 107 2nd pixel value generation part

Japanese Patent No. 4330164

Claims (9)

  In an image processing apparatus for generating a pixel value of an interpolation pixel in a predetermined image by a predetermined interpolation method, it is determined whether or not the image region including the interpolation pixel is a periodic region with periodic pixel value fluctuations. A determination unit, a first generation unit that generates a candidate pixel value of an interpolation pixel of the image using a first interpolation method, and a candidate pixel value of the interpolation pixel of the image using a second interpolation method When the second generation unit that performs the determination and the determination unit determines that it is a periodic region, the second generation unit selects a candidate pixel value generated by the first generation unit and determines that it is not a periodic region. Control means for selecting a candidate pixel value generated by the means; insertion means for inserting the selected candidate pixel value into the interpolation pixel; and a neighboring pixel existing in the vicinity of the candidate pixel value of the interpolation pixel and the interpolation pixel The pixel value of The image processing apparatus characterized by comprising updating means for updating using.   The updating unit calculates the correction amount using a pixel value of a local area including the interpolation pixel and a neighboring pixel and a pixel value of a peripheral area having the same size as the local area existing around the local area. The image processing apparatus according to claim 1, wherein:   The image processing apparatus according to claim 2, wherein the update unit calculates the correction amount based on an average pixel value of the local region and an average pixel value of the peripheral region.   The updating means detects a similar pattern similar to a reference pattern including the interpolation pixel by pattern matching, and calculates a difference between an average pixel value of the reference pattern and an average pixel value of the similar pattern as a correction amount. The image processing apparatus according to claim 2, wherein the image processing apparatus is an image processing apparatus.   The image processing apparatus according to claim 3, wherein the local area is the reference pattern, and the peripheral area is the similar pattern.   The updating means updates the candidate pixel value of the interpolation pixel and the pixel value of the neighboring pixel using a value obtained by superimposing a predetermined weight on the correction amount, and the weight is different between the interpolation pixel and the neighboring pixel. The image processing apparatus according to claim 1, wherein:   In an image processing method for generating a pixel value of an interpolation pixel in a predetermined image by a predetermined interpolation method, it is determined whether or not the image region including the interpolation pixel is a periodic region with periodic pixel value fluctuation A determination step; a first generation step of generating a candidate pixel value of an interpolation pixel of the image using a first interpolation method; and a generation of a candidate pixel value of the interpolation pixel of the image using a second interpolation method When the second generation step to be performed and the determination step are determined to be a periodic region, a candidate pixel value generated by the first generation step is selected, and when it is determined not to be a periodic region, the second generation step A control step of selecting candidate pixel values generated by the step, an insertion step of inserting the selected candidate pixel value into the interpolation pixel, and a neighboring pixel existing in the vicinity of the candidate pixel value of the interpolation pixel and the interpolation pixel The pixel value of An image processing method characterized by comprising the updating step of updating using.   A program for causing a computer to realize the image processing method according to claim 7.   A computer-readable recording medium having recorded thereon a program for causing a computer to implement the image processing method according to claim 7.

Images